Advertiser competition affects quality score
In response to this well written post by Apollo SEM “Visible Quality Scores are Similar to TBPR!“, I would like to clear up a couple of factors in his equation that nullify his findings. Though his blog is one of my favorites as a highly useful and unique article database, with articles such as: “NUDE: Create Dynamic Landing Pages In 15 Minutes Or Less” and “Getting Conversion Data With Your New Analytics Profile“, this last particular article is irrelevant. He makes many assumptions about quality score using his dynamically generated landing page theory. The problem is that he uses “homestar poopsmith” as the test keyword.
Click Photo to Enlarge
Quite simply, this is far too obscure of a word to make any true assessment of the effectiveness of his dynamic landing page technique. Try this technique on the keyword “mortgage” and see if you still get a minimum bid of 4 cents.
Here is an example of an obscure keyword that we use and get away with a 4 cent bid, without mentioning the keyword in the adtext: hosted exchange. Other than Bay Area, hosted exchange is not mentioned anywhere else.
The point is that advertiser competition has a huge impact on quality score. You can do everything right and still not get a great quality score if the term is competitive and the difference is CTR.
So let’s talk about CTR. Notice that the “homestar poopsmith” keyword has two impressions and two clicks. This is without doubt 2 clicks from Apollo SEM. This provides a perfect CTR which makes this test keyword inaccurate as well. Not only is the word obscure, the CTR is fraudulently perfect. Since he has launched his blog article he will now have random readers search for this keyword and click on the ad to see the example of the Dynamic landing page technique.
I think that it is a great technique but the test keyword makes all the findings irrelevant. Apollo SEM should put his money where his mouth is and use the technique on a highly competitive keyword. I am fully confident that the method will work and my hats off to them for figuring it out in the first place. My only concern is why they would use such an irrelevant method of testing when they obviously demonstrate a great PPC knowledge.
As for his ultimate conclusions from the test:
“… Visible Quality Score is similar to Visible (Tool Bar) PageRank - it really doesn’t mean anything. It is simply a visible indicator that can give you a vague idea of how relevant your keywords are.”
His test tries to find a relation between Quality Score and actual cost per click. Upon finding that his actual cost per click jumped around while quality score remained the same, he concludes that Quality Score “doesn’t mean anything”. We all know this is not true.
If you have a broadly matched keyword, of course the cost per clicks will jump around depending on what the actual query was, because using broad match is like bidding on multiple different keywords at once. However, you can’t extend this statement to conclude that Quality Score doesn’t matter to cost per click. Quality score does affect your cost per click, because it affects what your minimum bid must be for your ads to appear. If your ad can’t appear for 4 cents a click, you obviously have to pay more.
January 30th, 2008 at 11:49 am
Alex, I don’t exactly get your point with this post?
You start out trying to attack my theory that Quality Score is similar to TBPR…but you end with trying to stiffle the idea that dynamic landing pages aren’t the only thing that affect quality score - and you’re right about that. There are quite a variety of factors that affect quality score.
Google applied that QS min bid of .04 before any searches were even conducted on that term. We also bid $1 to test our theories…not .04.
The screen shot you use is not and was not meant to illustrate CTR…but using your theory that CTR and competition matter regarding click costs - you can note that at the time we had zero competitors on that keyword and you can see in the image that we did have 100% CTR but our cost was not even close to our min bid of .04…it was at .17 - nearly 4 x what the minimum bid/QS indicated the cost per click should be.
So having a very high CTR did not really affect our click cost…in fact, it made it more expensive per click from when we first initiated the test and we DID get a .04 cent click charged to us.
You certainly got my attention Alex.
Perhaps you can clarify your position though because I’m having a hard time relating your post title to the first half and last half of your post…?
Thanks.
January 30th, 2008 at 3:53 pm
Jameszol, Now that I have your attention
…
Many times articles inspire ideas that they did not intend to. This cascading of thought leads to healthy growth in the community. Your article brought up a great point of advertiser competition. I know your article was not actually about advertiser competition. It is my understanding that you were saying that quality score is a visual reference but is dynamic when it comes to the actual cost per click.
To be even more confusing, this last article that I commented on is actually stemming from you dynamic keyword insertion landing page article. So many references for the cascading to bounce off of, things can get a little confusing.
My only point that I would like you to consider for now is that your min bid (before the keyword ever goes live) in most cases will be much higher if it is not an obscure one. It is comparing apples to oranges. It is also much harder to get a great quality score with a highly competitive term.
I completely agree with you on the idea that quality score is determined by many different factors. The art is dissecting every factor one at a time. For example, mentioning your keyword in the headline of the text ad is a great practice. Mentioning that same keyword in the landing page is another great practice (thanks to you we can do so quite easily). But not every keyword is equally scrutinized. From a business standpoint, if there is a competitive keyword, Google is much more concerned with the bang for their buck.
“In many cases it takes Google up to 2 weeks to calibrate your quality score for new keywords [that is, correctly calibrate from the initial quality score].” - Google Rep
So how do you know if something is competitive or not (other than doing an obvious test search)? You follow all of Google’s best practices and if the keyword’s quality score is great, then it probably is not very competitive. If it is Poor or even Ok, then it is very likely that the keyword is very competitive.
Of course there is the historical quality score of that Ad Group, Campaign, and Account that take a big tole on the Pre-Live quality score of a new keyword.
January 30th, 2008 at 5:43 pm
A few things you took out of context in your post:
I almost always maintain that my little fun tests are not an adequate sample and everybody might experience something different. Most marketers have different organization methods and understanding/management styles. I am fairly clear about that in most of my posts.
I also noted that my observation regarding QS and TBPR was simply my opinion - and I’ll maintain that it is true. A broadly matched general keyword’s quality score is not representative of the broad match algorithm/history that Google maintains. What I mean is that a variety of words/phrases (we know that 25% of them are queries that Google hasn’t ever experienced) have to be dynamically and instantly related to the one general broad match term that was chosen in the account and the quality score for that exact query has to be dynamic to determine the relative cost to the account. I guess what that means is that the broader Google’s algorithm gets, the more they will charge for words that are further from the exact term chosen as a broadly matched term because the quality score dramatically lowers and therefore the dynamic min bid is higher - and that puts more money in their pocket so it makes sense for them to generalize the broad match as much as realistically possible.
I think what you are proposing is something that doesn’t necessarily disprove my theory/opinion but you are conducting an entirely different experiment. I was not measuring CTR - perhaps that was one control…what I wanted to do was see what would happen when the only real variable would be the landing page - scientifically, I had to choose an obscure word with zero competition to maintain control of the study. I had to ensure that the CTR would be high as that was part of the control. Competition would skew the study because your explanation of competition affecting QS is spot on. The ad was well designed to include the keyword too - another controlled part of the experiment. The only variable I wanted to manipulate was the landing page so scientifically speaking, my method feels like it was sound.
Now, your experiment and thoughts regarding CTR are true - and again, perhaps I should discuss my reasoning and controls in a future post so it doesn’t get confusing next time.
I appreciate a lively debate, thank you for posting this - I just subscribed to your blog.